Depth Image Rendering Splat Blending via Selective Offset
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional depth image-based rendering methods face challenges with slow blending speed using two paths, leading to image degradation and quality issues, especially as image resolutions increase, while methods using one path improve speed but compromise image quality due to improper offset determination.
Innovation Solution
A depth image-based rendering method that uses only one path for blending splats within a certain offset range, determining the rendering order based on depth information values, and selectively blending splats based on an offset value to enhance blending speed and image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If two paths are used for blending splats with z-offset inspection, then image quality is improved, but blending speed becomes slow
Solution Approach 1:
The patent extracts and eliminates the second path (z-offset inspection path) from the conventional two-path blending process. By using only the first path (McMillan algorithm path) with modified blending logic, the patent removes the redundant depth inspection step while maintaining image quality through selective blending based on depth information available during the single-pass rendering process.
Solution Approach 2:
The patent segments the blending decision process into depth-based selective blending within the single path. Instead of using a separate second path for z-offset inspection, the patent divides the blending operation to apply splats selectively based on depth comparisons performed during the first path's rendering sequence, thereby maintaining quality control without the overhead of a second traversal.
2Productivity
If one path is used for blending splats with McMillan algorithm, then blending speed is improved, but image quality degrades due to improper offset determination
Solution Approach 1:
The patent performs preliminary depth information preparation during the first path's rendering process. By computing and storing depth information for each splat as it is generated in the single pass, the patent enables subsequent selective blending decisions to be made immediately without requiring a second path for z-offset inspection, thus maintaining both speed and quality.
Solution Approach 2:
The patent implements feedback-based selective blending where depth information from the first path is fed back into the blending decision process. The system uses the depth values computed during rendering to determine which splats should be blended and which should be discarded, creating a closed-loop control that maintains image quality without requiring additional rendering passes.
Data Source
AI summary
A method, medium, and system of rendering image data, including receiving three-dimensional space data for each pixel forming the image data, the three-dimensional space data including depth and color information values for each pixel, determining an order of rendering for a plurality of surfaces forming the image data in an order of the distance between a rendering viewpoint and the surface forming the image data, generating splat depth information values for each splat included in the surface forming the image data, according to the rendering order, and determining whether the splat included in the surface forming the image data is required to be blended and accordingly selectively performing a blending on the splat.